Cell Inside, the Future of Processor Architecture - The Future of Cell and x86
(Page 6 of 7 )
The Sony-Toshiba-IBM trio expects that Cell will be able to be used in mobile devices, such as laptops and even cell phones. As of now, that’s not possible because the power consumption is too high. Cells clocked at 4 GHz reportedly suck down 80 Watts, not that bad for PC hardware but still too much to plant in a phone or PDA.
The companies have announced no plans to put the processor in mainstream consumer or business computers, though the idea wouldn't seem beyond Cell's scope. Once Cells are developed to be fully cooperative, the possibilities for the hardware are limitless. When buying a new computer, you won’t be replacing your old one; you’ll be supplementing it. Instead of computers being completely separate (and often useless after you replace one), old computers could still be very useful for donating extra processing cycles to the new system. Actually, with Cells going into TVs and game consoles, you could ideally network all your Cell-powered electronics and they could share and balance the workload between them.
So you're getting low FPS in some cutting edge game, say Doom 6 or Half Life 14; give a little boost by connecting your computer to your home entertainment system, microwave, and garage door opener. It can send some extra chucks of data out to these other devices for them to crunch. It will be like having your own makeshift computer array at home. Some say that Cell is only a media device, that it is useful in certain applications and is not useful for general computing. This really has absolutely no grounding yet, and it remains to be seen. The way programs are handled on Cell will not be the same as on current processors. They process "cells" of data and not a linear thread. If a stripped down Cell can render games smoothly on a PS3, and a full fledged Cell can clock up to 5.6 GHz already, the PC may have to get ready for a little competition.
Of course, PCs have already crushed a lot hardware that outperformed them in the past (anyone remember Amiga?). Even if Cell computers are produced and are far superior, the marketplace may not go their way. The key to Cell's success is probably marketting its interconnectivity and also starting outside the general consumer market, like in Mercury's products.
AMD and Intel must be ready for this new technology. The chip makers just released 64-bit CPUs and dual-core CPUs (to a market that isn’t quite ready), which seem to be born from a similar vision. PC manufacturers are already beginning to use the multiprocessor principles to accelerate those processes that will benefit from it. Cell's structure seems to be pointing to the future of CPU architecture, even in PCs.
However, software developers don’t seem ready to write the new software required for dual-cores, whether it’s because they aren’t comfortable with it or they don’t see enough advantage yet. Needing a new software base to properly take advantage of dual-cores certainly doesn't put the traditional PC hardware at any advantage. It also doesn’t help that those dual-cores only perform at a fraction of the capacity of a Cell. Though a poor excuse so far, they may be just the beginning. Before the Cell is practical outside of TVs and game consoles, they may look quite a bit more functional.
Next: It's the Software >>
More Computer Processors Articles
More By Developer Shed
/ 216 
